Image Transmission under Arche Conditions By Waldemar H. Lehn and H. Leonard Sawatzky'

Summary: The mirage, an optical effect fairly common in the high latitudes, occurs when the exhibits qr eater-then-norrnul refractive capability. Light rays then follow paths concave toward th e , with curvature qr eater than or equal to that of the earth. The atmosphere acquires this refractive capability under conditions of temperature , which exaggerate the fall-off of density (and hence ) with elevation. Equations are presented that ca1culate density profile from any given temperature p rofi le : these results arc used to develop ru y path equations for nearly horizontal rays in the lower atmosphere. Ray paths are computed for two distinct mirage conditions (a strong sh al low inversion and a mild but deep inversion) as weIl as for th e standard atmosphere. With the aid of these paths, the appearance of th e environment is shown. The strong inversion produces the appearance of a seucer-shapcd earth (with image dtstortion near the ), whereas the mild inversion produces a "flat e arth" appearance that permits great viewing distances.

Zusammenfassung: Die arktische ist ein verhältnismäßig häufig vorkommender optischer Effekt in höheren Breiten, Er entsteht, wenn durch gesteigerte Densitätsschichtung der Atmosphäre Lichtstrahl­ bahnen gebrochen werden in einem Ausmaß, das der Krümmung der Erde gleicht oder sie überschreitet. Die Atmosphäre gewinnt diese Refraktionsfähigkeit unter Temperaturinversionsverhältnissen, welche die normale Densi tätsabnahrna mit zunehmender Höhe verstärken. Gleichungen, die das Densitätsprofil bei gegebenem Temperaturprofil errechnen lassen, werden vorgelegt. Aus den Ergebnissen lassen sich Behn-Gleidiunqen annähernd horizontaler Lichtstrahlen im bodennahen Raum entwickeln. Lichtstrahlbahnen werden als Beispiele berechnet und dargestellt, sowohl für eine flache, Jedoch starke, und eine weniger starke, aber tiefe Inversion, als auch bezüglich der normalen Atmosphäre. Mit Hilfe dieser Ergebnisse wird die optische Gestalt der Umwelt errechnet. Die starke Inversion ergibt den Aspekt einer schüsselförmigen Erde; die weniger starken den einer absolut flachen Erde. Beide Er­ scheinungen verursachen eine optische Ausdehnung des Blickfeldes bei weit zurÜckweichendem Horizont.

1. Introduclion The arctic mirage is an optical phenomenon that occurs fairly frequently in the high latitudes. The effect, also called "100m" or "superrefraction", occurs when the atmosphere refracts light rays more strongly than usual. Under these conditions ray paths become concave toward the earth, with radii of curvature less than or equal to the aarth's radius. Visual information can then be transmitted beyond the normal "horizon distance", and the surface takes on a flat or saucer-shaped appearance. This paper describes the phys­ ical basis for the mirage, and develops equations useful in predicting the appearance of the environment under mirage conditions. Even the normal atmosphere refracts rays mildly, due to the decreasing density (and hence refractive index) with altitude. The effect is accentuated to produce the arctic mirage when a warm rests upon a cold surface. The conductive cooling from below develops a temperature inversion, with a steeper-than-normal density gradient. This gradient in turn produces the stronger effects that create the arctic mirage. This refracting layer need not be in contact with the aarth's surface. The situation where warm air overlies cold air is equally effective, the necessary density gradient occuring in the boundary zone between the two layers. Hobbs reports that, in the high latitudes, the cold layer is frequently of the order of 1000 meters deep. It should be noted that the arctic mirage is fundamentally distinct from the fata morgana or desert mirage, which depends on the opposite physical effect, namely pronounced heating of the air in contact with the surface. The following sections develop the equations that permit calculation of ray paths when the temperature profile in the lower atmosphere is known.

• Prof. Waldemar H. Lehn, Iteculty of Engineering, University of Mani toba, Winnipeg. Man. R3T 2N2 (Canada). Dr. H. Leonard Sawatzky, Department of Georgraphy, University of Manitoba, Winnipeg, Man. R3T 2N2 (Canada).

120 2. Ray paths in the refracting atmosphere

The refractive effects of the atmosphere are weIl known. For visible light Bertram gives the refractive index as a function of density p (kq/rn'') by

6 n = 1 + (226 x 10- ) p (1) and the propagation velocity v by

v = ein (2) where c is the velocity of light in vacuum. In the following analysis the atmosphere is considered to be laterally homogeneous, so that the only spatial variation in density comes with elevation z above the e ar ths surface. Attention will be concentrated on rays whose paths remain close to the surface; an analogous development can be used to analyse refractive zones at higher elevations. For such an atmosphere over a flat earth, Bertram derives an expression for the radius of curvature r at an arbitrary point on the ray: sin 8(z) dv (3) -=~ dz where O(z) is the angle between the ray and the vertical and v(z) is the velocity of propagation, at elevation z. From Sne lls Law of Refraction, the term (sin O)/v is a constant for any given ray. Coordinates are chosen such that the ray is contained in the xz plane. An additional set of coordinates (see Fig. 1) facilitates further calculations: a new abscissa u is chosen tangent to the ray path at its starting point, arid a new ordinate w pointing downward in the plane of the ray. On the basis that ray curvature is very small in the atmosphere, some approximations may be made:

x = u sin 8

z = z, + U cos 8

1::/« 1. Hence in the uw coordinates the curvature expression becomes 1 &,1 dw 5 (4) r = [I + (%i)'t = du' which combines with (3) to give I dw sin 8 dv (5) r = du 2 = -v- dz . In the following paragraphs the path equation for nearly horizontal rays will be derived, as a function of propagation-velocity gradient dv/dz. As the altitude variation of such rays is smalI, a Taylor expansion of dv/dz is useful:

d v , " ( ) 1 ," (z z )2 + (6) dz = V o + Vo Z - Zo + 2vo - 0 ••• where the prime denotes differentiation with respect to z. Substitution of (6) into (5) yields

-,dw ---sin 80 ( vo+vo(z-zo)+'" 21vo(z-zo)'" 2) . (7) du Va

Here the starting-point values 0 0 , v., are used for the constant ratio (sin O)/v. Following Bertram, one can now include the effects of the earths curvature. For horizontal distances under 1000 km, the equation of the earth's surface is very closely

121 z=-x2/2R, where R is the radius of the earth, In (7L the ray elevation above the surface 2 will thus increase by the additional term x /2R, Secondly, the angle 00 between ray and local vertical will decrease by the amount x/R, Incorporation of these effects into (7) produces an equation valid over the spherical earth:

2 dw sin(Bo - xlR) ( r "x -: x' ) W = v0 v0 + v0 (z + 2R - zo) + 2 (z + 2R zor· (8)

This equation must now be re-written in terms of uw coordinates. For nearly horizontal rays,

x = U u' (9) and z - Zo + u cas Bo --. 2ro The last term in the z-equation represents the effect of ray curvature: ro is the radius of curvature at u = O. The ro term can be replaced using (5);

v~sin 80 2 z = Zo + U cos 80 - ---u . (10) 2vo Further, the sine function in (8) is expanded in a Taylor series, with ~o defined as the complement of 0 0 : x u ~o sin(8 0 - p) = sin Bo - R sin . (11)

Equations (9L (10L (11) are substituted into (8) to give the differential equation for the ray path:

W ~o) v~ ~) d 2 _ (sin 80 - *' sin [ , " (. .h _ sin 80 2 2 - V0 + v0 u sin '!'o 2 u + 2R du v: Vo 2 (12) V~,(. v~ sin 80 2 u )2] + '2 u sm ~o - 2v u + 2R . o This equation is integrated twice to produce the equation of the ray path in the uw coordinate system. In each coefficient of un, any terms at least 1000 times smaller than their neighbours have been neglected.

[~2 v~ ~o ~~v~ v~v~sin v~'sin2~oJ _ sin 80 sin 3 _ 80 4 w - V 2 u + 6 u + 12\2R 2v + --2--; u o o (13) 1 (V~'Sin ~o v~v~'sin ~osin 80) 5 v~' (v~2Sin 280 1 v~sin 80) 6] +----- u+- +------u2 20 2R Zv, 60 4v; 4R 2voR

The subscripts "0" refer to values at the origin of the uw coordinates. Typical orders of magnitude for the velocity derivatives are v'Q "'" 100 sec", v~ - 10 m'l sec", and v: _ I m'2sec".

3. Relation belween velocily gradienl and lemperalure profile The path equation (13) requires the derivatives of propagation velocity with respect to elevation. These derivatives may be calculated from a given temperature profile as folIows. A standard result in elementary physics is the equation for pressure in an atmosphere situated in a gravitational field: dp - = - gp(z) (14) dz

122 Temperature (·C) 15 20

z

z o

o

o o o o o'" ~ N N N N N N '"N

Density (kg/m')

Fig. 1: Choice of co ordinate systems. Fig. 2: Temperature and density profiles: ex­ Abb. 1: Wahl der Koordinatensysteme. ample of strang inversion. Abb. 2: Temperatur- und Dichteprofile: Beispiel einer starken Inversion. where p is the pressure at elevation z, and 9 is the acceleration of gravity (assumed constant over the range of elevations to be encountered). Further. representing the behaviour of air for small deviations from standard temperature and pressure by the ideal gas equation pV = nRT, one can write the density as ßp p =7 (15) where T is the temperature in degrces K and ß is a constant of proportionality. For the Standard Atmosphere of the Air Research and Development Cornmand (see Mizner et al.) the value of ß is about 3.49'10-3 MKS units. Table 1 summarises some pertinent values for this atmosphere.

Atmospheric pressure at eerths surface: Po ~ 1.013,10' n/m' Surface density of air at 15° C: Po ~ 1.225 kq/rn" Effective molecular weight of air: 29.0 Radius 01 th e earth: R ~ 6400 km Velocity of light in vacuum: c = 3'10s rn/sec.

Tab. 1: NumericaI values used in calculations. Tab. 1: Bei der Berechnung benutzte Werte.

Substitution of (15) into (14) produces an equation that can be integrated; the result is P(z) dp' _ _ [z dz' (16) jp,f p' - gß Jo T(z') . The solution for p (z) is thus [z dZ'] p(z) Po exp [-gß Jo T(z') (17)

123 which when substituted back into (15) permits the calculation of density P (z) when the temperature profile T(z) is known:

ßpo [ \Z (18) p(z) = T(z) exp -gß 0 T(z')dZ'] .

The remammg step, from density to propagation velocity, is straightforward. Equations (1) and (2) combine to give c v = (19) 1 + 0.000226 P which to an accuracy better than one part in a thousand is equivalent to

v = c (I - 0.000226 p) . (20) Differentiation of (20) yields the velocity derivatives in terms of the density derivatives. The path equations were solved by a numerical procedure, thedetails of which are outlined below ".

Fig. 3: Ray paths for conditions of stronq temperature Inverston. Ray elevation above the earth's surface is plotted against distance along the surface. The observer's eye is 3 meters above the surface. The numbers on the paths give ray angles, in degrees above the horizontal, at the observers station. Abb. 3: Lichtstrahlbahnen bei starker Temperaturinveraion . die Höhe der Strahlbahn über der Erdoberfläche ist gegenüber der linearen Entfernung vom Ausgangspunkt dargestellt. Das Auge des Beobachters ist in 3 m Höhe. Die Aufschriften geben den Winkel eines Strahles zur Horizontalebene (in 0) an, vom Standpunkt des Beobachters betrachtet.

") The computational procedure used to calculate the ray paths will be briefly outIined. First a density table (equation [18]) at 4-meter intervals is ca1culated from the temperature profile data. To find the veloctty derivatives at any elevation, a cubic polynornial is fitted to the neighbouring density points. Differentiation of this cubic (and use of equation [20]) gives the velocity derivatives. Ray path cornpu­ tations then proceed in uw coordinates according to (13) untiI the u" term exceeds 0.001 m, at which time a new uw system is chosen. This shift of origin must be done carefuIly; as the new z-axis is not exactIy parallel to the original one {eerth's curvature) a corresponding correction must be applied to the new value of (/)0. Ray elevation above the eerths surface (assumed Iree of irregularities) is found by re­ turning to xz coordinates and using the relation z= -x2/2R to represent the eerths surface. The whole procedure is repeated until the ray intersects the earth or goe5 beyond the ar e a of interest.

124 Elevation Temperature Elevation Donsitv m oe m kg/m'

0.0 15.37 0.0 1.226 0.8 15.8 4.0 1.217 2 16.4 8.0 1.211 8.0 18.7 12.0 1.206 16.0 20.4 16.0 1.203 24.0 21.34 20.0 1.200 24.0 1.198 Tab. 2: Temperature end density data for example of strang inversion. Tab. 2: Temperatur- und Dichtewerte für das Beispiel einer starken Inversion.

o o '7(!/j // / 0.0

-0.02 / >­ / 0: > uJ-,0 WO / "'~,,~--~~

o o

-0.05/ o c

o c I I 40.UO 50.00 80.00 JOO,OO lk[J.oo lWO.OO ]60.00 )30.00 HORIZONTRL DISTRNCF [MfTFRSl ~ 10'

Flg. 4: Ray paths for the standard atmosphere. The numbers tndicate angles (in degrees above horizontal) at the observers eye, which is 3 meters above the surface. Abb. 4: Bahnen der Lichtstrahlen in der genormten Atmosphäre: die Aufschriften geben den Winkel zur Horizontalebene (in 0) vom Standpunkt des Beobachters (3 m Höhe) an.

4. Appearanee oi environment under strang temperature inversion Strong inversions ean appear when the atmosphere is eonductively cooled from below, as oecurs in overnight radiation eooling of the ground, or when warm air lies over cold water. An example of strong inversion was chosen from measurement data collected in Lettau and Davidson, Fig. 2 and Table 2 present the associated temperature and density profiles. Lyons reeords inversions of similar intensity. With the observers eye at an elevation of 3 meters, the ray paths shown in Fig. 3 were eomputed. This figure permits one to reconstruct the appearance of the environment under these conditions. For corn­ parison purposes Fig. 4 shows the ray paths for the standard atmosphere. The strength

125 Ejev ar l on Angle

0.05°

a

b

Fig. 5: Aspect of a sailboat under conditions of strong inversion (a) and standard atmosphere (b}. The t ip of tbe mast i s 7 meters above tbe waterline, while th e deck is 0.6 meter above tbe water; length of the boat is 4.5 meters. The vertical scale gives angles subtended at the observers eye, which is 3 meters above the water surface. Abb. 5: Aspekte eines Segelbootes unter Verhältnissen einer starken Inversion (a) bzw. genormter Atrno­ sphäre (b): die Spitze des Mastes ragt 7 m über die Wasserfläche, das Deck 0,6 m. Die Länge des Boots ist 4,5 m. Die senkrechte Skala gibt die Winkel zwischen der Horizontalebene und dem Ausgangspunkt des Lichts tr ahls , vom Standpunkt des Beobachters in 3 m Höbe über der Wasserfläche aus gesehen.

Elevation Angle

0.10

0.05

0.00

-0.05

b

Fig. 6: Aspect of buildings on very flat ground under conditions of strang inversion (a) and standard atmosphere (b). The buildings, al l of the s atne dimensicns, are respectively 5, 10, and 15 km from the ob ser ver, wbose eye is 3 meters above the ground. The buildings are 5 meters w ide : the rooftap and e aves are respectively 5 and 3 meters above the ground. Abb. 6: Aspekte Von Gebäuden auf flachem Gelände unter Verhältnissen starker Inversion (a) bzw. ge~ normter Atmosphäre [b}: Die Gebäude, von genau gleichmäßiger Größe, stehen in 5, 10 und 15 km Entfer­ nung vom Beobachter, dessen Auge sich in 3 m Höhe befindet. Die Gebäude sind 5 m breit; die Giebel und Dachkanten ragen 5 m bzw. 3 m über die Erdoberfläche. of the effect seen by the observer is emphasized by the presence of familiar references in the field of view. To illustrate this, an over-water view of a small sailboat 5 km from the observer was choscn. The dotted vertical line on Fiq. 3 shows the location and size of the boat. The aspect of the boat is drawn to scale in Fig, 5, where Fig, 5a is for the conditions of strong inversion, and Fig. 5b for the standard atmosphere. The reference scale represents the angle subtended at the unaided eye. Note that in Fig. 5a the horizon

126 is clearly higher than the top of the mast, ieaving the viewer with the distinct impression that the earths surface is saucer-shaped. Fig. 6 shows a similar situation over extremely flat ground, as exemplified by the Lake Agassiz plain in southern Manitoba (the inhabitants of this area are familiar with these visual effects). The situation illustrated represents three villages at distances of 5, 10 and 15 km respeetively from the ob server. Under the mirage conditions (Fiq. 6a) the villages appear "staircased" one above the other, and the horizon is higher than all of them. An image distortion that increases with distance also be cornes apparent, due to compression of ray paths as the horizon is approached: rays originating at progressively farther points subtend progressively smaller angles at the observers eye. In the normal view, on the other hand, the first village stands much higher than the second, and the third barely rises over the horizon.

5. Condilions for viewing greal dislances For a line of sight of maximum length, the ray path must have a radius of curvature equal to that of the carth. For a surface temperature of 00 C, the requisite temperature gradient was found to be 0.112 0 C per meter. Fig, 7 shows the resulting ray paths, assuming the somewhat idealised case for which the inversion layer with this gradient has aSO-meter dcpth. As a specific example, an object 50 m high located 200 km from the observer subtends an angle of about 113 minute of arc. Under conditions of good illumination arid contrast, such an object is visible to the naked eye. In a stable inversion la y er of unlimited extent, the ultimate limit on viewing distance would be set by light absorption in the atmosphere, The appendix gives the formula for the appropriate transmission factor. Image transmission near the earth's surface is feasible up to 400 km, of sufficient quality that the unaided eye could derive useful information from it. If the image is transmitted at higher levels in the atmosphere, as occurs when the refracting layer is at the upper boundary of a deep cold-air layer, the image quality is improved: the thinner air at the upper levels absorbs less light, so that m or e of the light leaving the object reaches the observer. Stefansson reports a well­ authenticated, clear viewing of Snaefells Jökull in Western Iceland from a position at

/ / / j' / ~// //

, !.jUiJ.oo

Fig. 7: Ray paths that permit very large viewing distances. The nurnbers indicate angles (in degrees above horizontal) at the observers eye, 3 meters above the surface. Abb. 7: Bahnen der Liditstrahlen, die gesteigerte Sichtweite ermöglichen. Die Aufschriften geben die Win­ kel (in 0) zur Horizontalebene in 3 m Höhe an,

127 sea over 500 km distant; and Hobbs dtes numerous cases in which !ines of sight range from 300 to 400 km in length.

6. Conclusions The arctic mirage occurs when a temperature inversion produces a steepened density gradient in the lower atmosphere. The preceding paragraphs give equations that relate temperature profile to density, and density profile to (nearly horizontal) ray paths. With the aid of these equations, the appearance of the environment can be predicted for any lower-atmosphere temperature profile. The effects of the mirage are noticeable primarily over flat, relatively featureless terrain. For strong inversions near the surface the observer sees a saucer-shaped earth, with objects near the horizon somewhat distorted. On the other hand, viewing over very great distances is made possible by mild deep inversions, which when located near the surface produce the appearance of a flat earth. Historical and cultural implications of the arctic mirage are examined in aseparate paper. There the authors contend that the mirage may have had significant influence in the formation of the medieval world-image, and in early North Atlantic exploration. The mirage permitted the experienced but unsophisticated observer of the previous millenium to confirm theories of a flat or saucer-shaped world; in addition he could derive practical benefit from the mirage when it transmitted information from beyond the normal horizon.

References Bau r , F. : Meteorologisches Taschenbuch, VoL Ir. Leipzig, Akademische Verlagsgesellschaft, 1970. B e r t r a m, S.: Compensating for Propagation Errors in Electromagnetic Measuring Systems. I.E.E.E. Spectrum, March 1971. Hob b s, W. H. : Condilions 01 Exceptional Visibility wilhin High Latitudes, Particularly as a Result of Superior Mirage. Annals of the American Association of Geographers, December 1937. Let tau, H. H. an d B, Da v i d s 0 n (e d.) Exploring the First Mi le : Proceedings 01 the Great Plains Turbulence Field Program. New York, Symposium Publications Division, Pergamon Press, 1957.

LY 0 TI S r W. A. : Numerical Simulation of Great Lakes Summertime Conduction Inversions. Proc. 13th Conference on Great Lakes Resources, 1970. Mi z n er, R. A., K, S, W. C ha m p ion an d H. L. Po n d : The ARDC Model Atmosphere, 1959. Air Force Surveys in Geophysics, no. 115. Moor e, R. K.: Troposphertc Propagation Effects on Earth-Space Low Elevation Angle Paths. Lawrence, Kansas, U. of Kansas Publications, BuH. of Engineering No. 61, 1970.

R 0 s s i I B. : Op tics. Reading, Mass., Addison-Wesley Publishing Co" 1965. S tel ans s 0 n, V. : Ultima Thuje. New York, N. Y., The Macmillan Co" 1940. Appendix The extinction coefficient, a, for dry air is given in Baur: 8n 3 (n; - l)2ld a = 1.061 3 , N;Jc 4 oN dx where n, refractive index of air at S.T.P" N s number of air moleeules per cubic centimeter at S.T.P., N number of air moleeules per cubic centimeter along ray path, d distance of object from observer, t. wavelength of radiation.

The light-transmission factor q is q = exp (-a). As a specific example, consider a path of length 200 km at a small, roughly constant elevation. For air temperature 0° C and light wavelength 0.6,u, the extinction coefficient is a = 1.72, and the transmission factor is q"'" 0.18. Thus 18% of the light intensity originating at the object will reach the ob server. A similar calculation for air at 15° C leads to the values q c-: 9010 for a 300 km path length, and q "'" 4010 for a 400 km path.

128